This article is shared from HUAWEI CLOUD Community's "[HDC.2024] Cloud Native Middleware, Building a Secure and Trustworthy Connection Bridge for Software-Cloud Community-HUAWEI CLOUD".
Recently, during the HUAWEI CLOUD Developer Conference 2024, senior experts, technical directors, and senior product managers from the product teams in the field of HUAWEI CLOUD PaaS services and middleware delivered speeches on the theme of "Cloud-native middleware, building a secure and trustworthy connection bridge for software".
At the beginning of the speech, Long Ying, a senior expert of Huawei Cloud products, first analyzed the trend of middleware in the development of cloud computing, pointing out that from the end of the last century to the beginning of this century, the development scale of the software industry has become increasingly large, and the software industry chain has become more and more refined, and the complexity is no less than that of the supply chain system of the traditional manufacturing industry. As a result, the entire software encounters very big challenges, and it needs to be upgraded through costly means such as refactoring.
Long Ying, senior expert of HUAWEI CLOUD products
HUAWEI CLOUD middleware: provides state-of-the-art, secure, trusted, and diverse middleware to help customers build world-class software
With the technological development of cloud computing, cloud-native middleware, as a key component of cloud-native architecture, is also facing a series of challenges in many aspects such as core capacity building, business continuity, and ecological richness.
The construction progress does not meet the customer's requirements for cloud service capabilities. On the one hand, there is a mismatch between talent and organization and business development, especially the construction of talent echelon takes a long period and cost, which can be gradually improved for a large enterprise, but it is very difficult for a small company to achieve; On the other hand, the reliability problems caused by security and technical bottlenecks also make the operation of enterprises have great hidden dangers.
Business continuity risks associated with the use of open source middleware. According to the survey data, more than 95% of enterprises are using open source middleware products, and open source middleware products usually have some security vulnerabilities, which may be maliciously exploited, resulting in serious consequences such as system attacks and data leakage. Reports, such as Synopsys' Open Source Security and Risk Analysis Report, have a whopping 84% of their codebase containing at least one known security vulnerability that, if exploited by an attacker, could have a serious impact on an organization's business continuity.
Middleware comes in a variety of forms, and customers expect cloud vendors to be able to cover all aspects. In the face of many middleware products, customers hope to be able to achieve business migration to the cloud at the minimum cost, and the problem of unreasonable middleware selection in history can be solved at this stage.
In order to meet the above challenges, HUAWEI CLOUD Middleware has implemented different countermeasures to effectively solve customer business problems.
First of all, it is divided into "three steps" to match customers to the cloud, and it has now entered the second stage
Phase 1 focuses on enabling customers who move to the cloud to have a great experience with cloud capabilities. Currently, the OpenAPIs of almost all services on HUAWEI CLOUD are hosted on the cloud. For example, DCS can improve the concurrency performance of a single shard from 100,000 to 450,000 and improve the stability of delay jitter waveforms by 10%.
In the second stage, a large number of capabilities can be built that allow customers to use the cloud in depth, and a lot of effort has been invested in polishing the middleware product in terms of resilience and fast recovery. For example, the intelligent diagnosis capability of Distributed Message Service (DMS) can diagnose and provide optimal optimization suggestions with one click. Chaos engineering products support 10 major scenarios and 100+ failure modes to detect faults, helping customers minimize the hidden danger coefficient of using cloud services.
In the third stage, we build cross-cloud capabilities, build a complete set of solution products from the storage layer, data layer, and middleware layer, and carry out unified management of multi-cloud and multi-cluster through the large visual screen.
Second, HUAWEI CLOUD middleware has repeatedly hammered out excellent products through Huawei's management system
Through continuous investment, a very complete set of system standards has been established, which has been integrated into the company's R&D IPD and DSTE processes, realizing the transformation from relying on individuals and launching successful products by chance to institutionalizing and continuously launching high-quality products. Huawei's products and solutions have been operating safely and stably in more than 170 countries, and have earned the trust of many customers around the world.
In addition, we should deepen the industry, find technical partners for the scene, and establish a community of interests
To better meet customer needs and accelerate the digital transformation of industries, Huawei has set up a team organization for multiple industries, such as coal mining, ports, highways, electric power, and finance, to find matching technologies from the perspective of the industry. The legion is a professional team that provides end-to-end resources for the company, including R&D, sales, delivery, service, etc., in order to deeply understand the production practices and industry knowledge of the industry, and create products and solutions that adapt to the scene. Cooperate with more solution partners to integrate the capabilities of Huawei and partners to form targeted solutions. On the other hand, for the commercial market, product sales, delivery, and maintenance are almost all led by a large number of partners, and Huawei will adapt end-to-end from R&D, marketing, sales, supply, and service, enabling partners to independently expand and deliver Huawei products and solutions.
Based on the above solutions, we have finally created a complete set of middleware products, and adopted a two-wheel strategy of self-development + openness to support HUAWEI CLOUD's middleware product series to grow bigger and deeper. HUAWEI CLOUD Middleware is committed to providing technology-leading, secure, trusted, and diverse middleware components to help customers build world-class software. At present, it has been implemented in the world's hottest new energy industry, and has accumulated a wealth of practical cases.
The road of innovation and acceleration in the Internet of Vehicles industry is stable and far-reaching
At the meeting, Ming Ze, Technical Director of HUAWEI CLOUD Middleware, mentioned that with the increasing popularity of automobiles and the continuous development of technology, the trend of "New Four Modernizations" in the global automotive industry is becoming more and more obvious.
Ming Ze, Technical Director of HUAWEI CLOUD Middleware
Networking is the foundation of digitalization and servitization, and on the basis of a large amount of data, various intelligent services have emerged, especially autonomous driving, which has completely laid the core position of intelligence. Mobility services have made vehicle sharing a new mode of consumption. In addition, globalization is also a major trend, and car companies are expanding into new market patterns through the new energy track, and one of the basic requirements of this process is to meet the business security compliance of the host country. Mingze said that as the automotive industry enters the era of 100 million-level vehicle connectivity, cloud-native middleware has become a key technology to support massive data management and rapid software iteration, which not only optimizes intelligent driving services, but also promotes the digital transformation of car companies.
Middleware is divided into two scenarios in the intelligent construction of the Internet of Vehicles, among which the vehicle data reporting scenario, including vehicle condition (tire pressure, battery), POI location information (trajectory, peripheral), driving data (speed, acceleration, braking, idling, etc.), air conditioning, seat heating, entertainment data, etc., needs to be able to support tens of millions of vehicle-machine connections and tens of millions of topic topics; It supports powerful rule engine data integration, flexible delivery of rules, rapid collection, and million-level message throughput, and real-time alarm and event-driven. Another scenario, that is, remote control of the car through the mobile app, such as remote start, air conditioning, trunk, window lifting, lights, door, remote diagnosis, etc., needs to support device decoupling, reduce the difficulty of APP and TSP development, and support offline messages.
These complex scenarios can be effectively solved by the key capabilities of different middleware product portfolios, such as Distributed Cache Service (DCS), Distributed Message Service (DMS), Event Grid (EG), and Function Workflow (FG), and have been maturely applied in the business of many car companies.
New Cache Release: Breaking the boundary, the caching product form follows the continuous transformation of the business
Middleware is an indispensable basic component of a distributed system, and the data cache access acceleration service is an important part of middleware, which can support the millisecond-level task delay of tens of millions of concurrent tasks in the system and improve the stability and availability of the system.
Senior Product Manager of HUAWEI CLOUD Middleware, Meteor
In his speech, Lei Xing, Senior Product Manager of HUAWEI CLOUD Middleware, mentioned that compared with the self-built DCS for Redis, the basic performance of DCS is significantly better than that of the open source Redis, and it also provides many valuable features, such as one-click online specification change and 30+ monitoring items. Enterprise Edition provides capacity-based products, which can replace memory with SSDs to obtain significant cost advantages.
In terms of capacity expansion, the traditional expansion method is to double the capacity, such as the configuration of 3 nodes and 32 GB, and the expansion can only double the number of nodes or capacity, the step size is large, and the utilization efficiency is insufficient. In response to this kind of problem, Meteor said that DCS provides fine-grained elastic scaling, which increases the number of nodes linearly, and the performance also increases linearly, and the concentration of hot keys can also be quickly reduced through shard scaling.
In addition, DCS also provides an independent scale-out item-bandwidth, which allows you to increase the bandwidth limit of the shard by scaling up the bandwidth of a single shard to ensure stable and reliable performance when the traffic of a hot key in a single shard exceeds the bandwidth of a single shard or the bandwidth of an entire instance.
At the end of the speech, Meteor also mentioned that in the course of years of business practice, DCS has accumulated and summarized a wealth of high-availability disaster recovery scenarios, such as primary/standby replicas in AZs, single-region cross-AZs, and multi-region multi-AZs.
Distributed Messaging Service (DMS) is a new feature that comprehensively improves the architecture and provides new options for enterprise-level messaging middleware
DMS is currently positioned as the core key component of modern streaming architecture, providing cloud service capabilities compatible with Kafka, RocketMQ, and RabbitMQ, and is widely used in scenarios such as microservices, cloud native, big data, and finance, and is favored by users in business systems such as Internet of Vehicles, user behavior analysis, feed stream recommendation, and e-commerce.
Dapeng, Senior Product Manager of HUAWEI CLOUD Middleware
In this speech, Dapeng, Senior Product Manager of HUAWEI CLOUD Middleware, first explained the advantages of DMS:
- Stable and reliable, providing 95% availability, solving the problem of open source availability, and ensuring the stable operation of customer business.
- Extreme performance and elasticity, it supports tens of millions of concurrent and trillions of messages at most, and has stronger transaction processing and scheduled message capabilities.
- In terms of security, it provides a full range of security protection from data transmission, storage security, sensitive operations, etc.
In the speech, Dapeng also brought a new product series to be launched: a new version of Kafka, a new version of RocketMQ, and a new version of RabbitMQ.
- A new version of Kafka
It will provide new serverless specifications and the combination of basic bandwidth + elastic bandwidth, with a maximum of 1 GB of elastic bandwidth, bringing customers a new Kafka experience and better service adaptability.
- A new version of RocketMQ
The new Basic and Professional product lines will be available, which will significantly reduce the starting specifications while optimizing the storage software and hardware architecture, reducing the overall storage cost by more than 30%. In addition, the new Professional Edition instances will also support up to 50,000 elastic TPS capabilities, which can better adapt to customers' business fluctuations in different scenarios.
- A new version of RabbitMQ
While compatible with the open-source AMQP protocol, the professional version of this release also solves the stability and reliability problems of the open-source kernel, including the high watermark and stop response problems faced by the open-source version, providing customers with better product choices for their business.
EventGrid: Best practices for breaking through event silos and using event streams to implement data routing
EventGrid is a serverless event bus service provided by HUAWEI CLOUD that can flexibly route events between applications using the standardized CloudEvents protocol, helping users easily build loosely coupled, distributed, and event-driven serverless microservice applications.
At the meeting, Kang Kang, Senior Product Manager of HUAWEI CLOUD Middleware, gave a detailed introduction to the three advantages and capabilities of the service:
Kang Kang, Senior Product Manager of HUAWEI CLOUD Middleware
- Event Center has 100+ built-in HUAWEI CLOUD event sources and supports customization and partner event extensions
Massive official data sources are supported, covering databases, messages, function computing, big data, DevOps platforms, IOT, etc., and custom event integration is supported.
- Interface: Easy integration and online processing of events
Interface-based, easy integration and online processing of events, simplified event integration, click to configure event sources and targets; It supports online real-time change event filtering, conversion and other processing logic, as well as custom complex format conversion.
- Event O&M and observability
The O&M capability is manifested in three aspects: event query, event trajectory, and event debugging. The observability mainly refers to 15+ monitoring indicators in operation, success, number of failures, and time-consuming monitoring, and supports a large monitoring screen, which is convenient for the observation system and operation logs to be displayed one by one, and users can quickly locate problems.
In addition, regarding the four typical scenarios of EventGrid, Kangkang also said that in the Internet of Vehicles solution, EventGrid plays the role of event center, event driven, and streaming ETL data cleaning. In the data routing synchronization scenario, as a data synchronization pipeline, it connects different systems and services, supports arbitrary communication of global events, and cloud messages, cache routing, synchronization, backup, and cross-region and off-site multi-active, helping customers easily realize migration to cloud data synchronization, cross-region backup disaster recovery, account data synchronization and cross-IDC hybrid deployment data synchronization, and create an integrated message solution. This is a case in which multiple active data centers in two cities and three data centers are synchronized in real time based on EventGrid event streams.
At the technical architecture level, the HUAWEI CLOUD middleware product team hopes to build a unified architecture in terms of control, scheduling, O&M, and authentication, and provide customers with more efficient, stable, and secure cloud-native services through continuous technological innovation and service upgrades. In the future, HUAWEI CLOUD will continue to explore industry scenarios and work with customers and partners to make more innovative middleware products in the joint construction of infrastructure.
Follow #HUAWEI CLOUD Developer Alliance# Click below to learn about HUAWEI CLOUD's fresh technologies~
HUAWEI CLOUD Blog_Big Data Blog_AI Blog_Cloud Computing Blog_Developer Center-HUAWEI CLOUD